Molecular Mechanisms of the Combination of Retinoid and Interferon-gamma for Inducing Differentiation and Increasing Apoptosis in Human Glioblastoma T98G and U87MG Cells

Abstract

Glioblastoma is the deadliest brain tumor that remains incurable. We examined efficacy of combination of retinoid and interferon-gamma (IFN-γ) in human glioblastoma T98G and U87MG cells. We conjectured that retinoid could induce differentiation with down regulation of telomerase activity to increase sensitivity to IFN-γ for apoptosis in glioblastoma cells. Indeed, treatment of cells with 1 μM all-trans retinoic acid (ATRA) or 1 μM 13-cis retinoic acid (13-CRA) for 7 days induced astrocytic differentiation with upregulation of glial fibrillary acidic protein (GFAP) and down regulation of telomerase activity. Wright staining and ApopTag assay showed, respectively, morphological and biochemical features of apoptosis in glioblastoma cells following exposure to 200 units/ml IFN-γ for 48 h. Induction of differentiation was associated with decreases in levels of nuclear factor kappa B (NFκB), inducible nitric oxide synthase (iNOS), and production of nitric oxide (NO) so as to increase sensitivity to IFN-γ for apoptosis. Notably, IFN-γ induced signal transducer and activator of transcription-1 (STAT-1) to bind to gamma-activated sequence (GAS) of the target gene. Also, IFN-γ activated caspase-8 and cleaved Bid to truncated Bid (tBid) for translocation to mitochondria. Fura-2 assay showed increases in intracellular free [Ca²⁺] and activation of calpain in apoptotic cells. Besides, increases in Bax:Bcl-2 ratio and mitochondrial release of cytochrome c and Smac into the cytosol activated caspase-9 and caspase-3 for apoptosis. Taken together, our results showed that retinoid induced astrocytic differentiation with down regulation of telomerase activity and enhanced sensitivity to IFN-γ for increasing apoptosis in human glioblastoma cells.

Introduction

Glioblastoma continues to be associated with a dismal prognosis, despite aggressive therapy [1]. Poor prognosis of glioblastoma patients has not changed over the years. Because traditional treatment strategies are ineffective, use of innovative therapeutic strategies is essential for treating this deadly disease. In particular, multimodality treatment with different chemotherapeutic agents improve outcome and currently combination chemotherapeutic options are showing increased impact in the overall outcome [2]. Several cytotoxic chemotherapeutic agents that are efficacious in treating glioblastoma are currently in clinical use. This investigation was carried out to develop a dual approach to control the growth of glioblastoma by promoting differentiation and enhancing apoptosis.

Since retinoids inhibit proliferation and induce differentiation as well as apoptosis in hepatic, pulmonary, colorectal, and other cancer cell lines [3, 4], we have studied the effects of all-trans retinoic acid (ATRA) and 13-cis retinoic acid (13-CRA) on two glioblastoma cell lines. ATRA is a biologically active retinoid whereas 13-CRA is a stereoisomer of ATRA and has shown anti-tumor activity in the treatment of hematologic and solid malignancies [5]. Some of these retinoids are already in clinical use. In some cases, retinoids were more active in combination with another therapy, for example, more apoptosis occurred in cancer cells with ATRA plus cisplatin [3] and ATRA plus IL-2 [4] than each drug alone. A large body of preclinical and clinical data indicated the potential of combination of a retinoid and an interferon (IFN). The combination of ATRA plus IFN-α reduced cancer cell proliferation by 50–60% [5]. ATRA exhibits anti-proliferative properties, independent of terminal maturation, through down regulation of telomerase leading to telomere shortening [6]. Combination of 13-CRA and IFN-α−2A is effective in the treatment of specific epithelial malignancies that include squamous cell cancers of the skin or cervix, and advanced renal cancer [7]. Also, combination of 13-CRA and taxol, cisplatin, and/or IFN-α entered Phase I/II trials for the treatment of head and neck squamous cell carcinomas, non-small cell lung carcinomas, and prostate cancer [8, 9]. Clinical and experimental models have shown that combination of ATRA and IFN-γ (henceforth ATRA/IFN) potently inhibits tumor cell growth and angiogenesis [10], compared to either agent alone. The retinoids and interferons (IFNs) are naturally occurring substances, which have strong anti-proliferative actions against breast cancer cells and, combination of a retinoid and an interferon, act additively or synergistically to inhibit growth of breast cancer cells in vitro and in vivo [11]. The growth inhibitory action was much more extensive when the two drugs were used together. These observations prompted us to examine the efficacy of ATRA/IFN in controlling the growth of human glioblastoma T98G and U87MG cells.

The extrinsic pathway of apoptosis is initiated by activation of the apical caspase-8 following death receptor ligation. Active caspase-8 processes Bid to truncated Bid (tBid), the C-terminal part of Bid, which is then translocated to mitochondrial membrane to trigger cytochrome c release and activation of caspase-9 [12–14]. On the other hand, drug-induced stress initiates the intrinsic pathway of apoptosis to activate apical caspase-9 directly. These pathways converge with activation of the executioner caspase-3. Superimposed on this scheme is an “integration model” in which both upstream and downstream caspases as well as other proteases cooperate for mediating cell-specific apoptosis.

The IFNs are introduced as the cytokine therapeutics into clinics for treating more than a dozen of viral diseases and also various malignancies such as chronic myelogenous leukemia, glioblastoma, hairy cell leukemia, Kaposi’s sarcoma, melanoma, multiple myeloma, ovarian cancer, and renal cell carcinoma [11]. At low doses, IFNs promote differentiation and show anti-angiogenic activity along with stimulation of the immune system by T-cell proliferation and upregulation of antigens [15]. However, the complete anti-tumor activity of IFNs is not understood completely. It has been reported that IFN-γ induces apoptosis in glioblastomas [16]. Glioblastoma cells express a similar set of genes for the receptors of cytokines, including IFN-γ [17]. Treatment of glioblastoma cells with IFN-γ decreased cell viability, proliferation, and also invasiveness [18]. Our laboratory also reported that IFN-γ upregulated calpain for mediation of apoptosis [19]. Some studies suggest that IFNs activate caspases. Induction of apoptosis by IFN-γ requires activation of JAK-1 and STAT-1 and also an increase in expression of caspase-1 [20].

Our studies indicate that treatment with a retinoid (ATRA or 13-CRA) induces astrocytic differentiation with down regulation of telomerase activity in T98G and U87MG cells and increases sensitivity to IFN-γ for apoptotic death via multiple pathways.

Materials and Methods

Cell Culture

Human (T98G and U87MG) glioblastoma cells were grown in 75-cm² flasks containing 10 ml of RPMI 1640 (Sigma, St. Louis, MO, USA), 1% penicillin and streptomycin (GIBCO, Grand Island, NY, USA), and 10% FBS (GIBCO). ATRA and 13-CRA (Sigma) were dissolved in dimethyl sulfoxide (DMSO) and stored as aliquots of 1000× stocks at −70°C. Since the retinoids are sensitive to light, all experiments involving treatment with retinoids were performed under subdued lighting. Each experiment included control cultures that received the same volume of DMSO solvent that was administered in the retinoid treatment. The concentration of DMSO in each experiment was always less than or equal to 0.01%, which was not toxic and did not induce differentiation. Optimum doses of IFN-γ (Calbiochem, San Diego, CA, USA) were determined for induction of apoptosis in glioblastoma cells. Cells were treated with IFN-γ alone for 48 h. For all incubations, cells were kept in a 37°C incubator with 100% humidity and 5% CO₂.

Methylene Blue Staining for Examination of Morphological Changes due to Induction of Astrocytic Differentiation

Human glioblastoma T98G and U87MG cells were cultured in monolayer in 9-cm diameter plates in absence and presence of ATRA or 13-CRA. Culture medium was aspirated and washed with ice-cold phosphate-buffered saline (PBS, pH 7.4) two times. Then, each plate was placed on ice and 5 ml of ice-cold 50% (v/v) ethanol was added to fix the cells. Ethanol was aspirated followed by the addition of 5 ml of ice-cold 0.2% (w/v) methylene blue solution (made up in 50% ethanol). The cells were stained for 30 s and washed twice with ice-cold water. The plates were then dried in the air and the cells were examined under light microscopy at 400× magnification for induction of astrocytic differentiation.

Detection of Morphological and Biochemical Features of Apoptosis

Human glioblastoma T98G and U87MG cells from each treatment were detached with a cell scraper to harvest attached and detached cells together. Cells were washed twice in PBS and sedimented onto the microscopic slide using an Eppendorf 5804R centrifuge (Brinkmann Instruments, Westbury, NY, USA) at 106g for 5 min. Cells were fixed for Wright staining, as we described previously [12–14]. Morphological features of apoptosis were examined under light microscopy. Cells were considered apoptotic if they showed (i) reduction in cell volume and (ii) condensation of the chromatin and/or the presence of cell membrane blebbing.

For in situ detection of DNA fragmentation in apoptotic cells, we used ApopTag assay kit (Intergen, Purchase, NY, USA. Apoptosis was assessed by counting ApopTag-positive cells on the grid of the microscopic field at 400× magnification. The counterstain methyl green stained normal nuclei a pale to medium green. The nuclei with DNA fragments were stained dark brown by the ApopTag detection system and were not stained with the methyl green. Experiments were performed in triplicate.

Detection of Telomerase Activity

Telomerase activity was assayed by application of the Telomerase Repeat Amplification Protocol (TRAP) [21] with some modifications. Briefly, cells were washed once with ice-cold PBS, again with ice-cold wash buffer, lyzed in ice-cold lysis buffer for 30 min on ice, and centrifuged in a microfuge at 4°C. The supernatant was removed and quick frozen on dry ice. The protein concentration was measured by Coomassie Brilliant Blue assay. For the TRAP assay, we used 150 ng protein per reaction. Assay tubes were prepared by sequestering 0.2 μg CX primer (5′-CC CTT ACC CTT ACC CTT A CC CTT AA-3′) under a wax barrier known as Ampliwax (Applied Biosystems, Foster City, CA, USA). Each extract was assayed in 50 μl of reaction mixture containing 20 mM Tris–HCl, pH 8.3, 1.5 mM MgCl₂, 63 mM KCl, 1 mM EGTA, 0.1 mg/ml BSA, 0.005% (v/v) Tween-20, 1 ag ITAS, 0.2 μg TS primer (5′-AA TCC GTC GAG CAG AGT T-3′), 1 μg T4g32 protein (Roche Applied Science, Indianapolis, IN, USA), and 2.5 units Taq DNA polymerase. After 30 min of incubation at 23°C for telomerase-mediated extension of the TS primer, the reaction mixture was heated at 90°C for 3 min and then subjected to 35 cycles (94°C for 45 s, 50°C for 45 s, and 72°C for 1 min) of polymerase chain reaction (PCR). The PCR products were subjected to electrophoresis to resolve on 12% polyacrylamide gels. The gels were stained with ethidium bromide (1 μg/ml), examined under UV (303 nm) light, and photographed.

Determination of Intracellular Free [Ca²⁺] Using Fura-2 Assay

We used the fluorescence Ca²⁺ indicator fura-2/AM, as we described previously [12–14], for determination of levels of intracellular free [Ca²⁺] in T98G and U87MG cells. The intracellular free [Ca²⁺] was calculated spectrofluorometrically using the equation [Ca²⁺] = K_dβ (R – R_min)/(R_max − R) [β, is the ratio of ³⁸⁰F_max (fluorescence intensity exciting at 380 nm for zero free Ca²⁺) to ³⁸⁰F_min (fluorescence intensity exciting at 380 nm for saturating free Ca²⁺)]. The determination of fluorescence ratio (R) was performed using an SLM 8000 spectrofluorimeter (Thermospectronic, Rochester, NY, USA) at 340 and 380 nm wavelengths. The maximal (R_max) and minimal (R_min) ratios were determined using 200 μl of 250 μM digitonin (Fisher Scientific) and 500 mM EGTA (Sigma), respectively, in 1 ml solution. The cell-specific dissociation constant (K_d) for T98G was 0.387 μM and for U87MG was0.476 μM, as we determined using standards of the Calcium Calibration Buffer kit with Magnesium (Molecular Probes, Eugene, OR, USA).

Electrophoretic Mobility Shift Assay (EMSA)

We used a standard method to prepare the nuclear protein samples for testing the activation of specific DNA-binding site, explicitly, gamma-activated sequence/IFN-γ-stimulated regulatory element (GAS/ISRE) [22]. Reactions for nuclear protein-DNA binding were performed using the Light Shift Chemiluminescent EMSA kit (Pierce, Rockford, IL, USA). The nucleotide sequences (forward: 5′-AAG TAC TTT CAG TTT CAT ATT ACT CTA-3′ and reverse: 5′-TAG AGT AAT ATG AAA CTG AAA GTA C-3′) corresponding to the STAT-1 consensus DNA-binding site, GAS/ISRE, were obtained from Qiagen (Fremont, CA, USA). Reaction mixtures were incubated at room temperature for 30 min and then resolved on 4% polyacrylamide gels using the running buffer 25 mM Tris–HCl, pH 8.3, 190 mM glycine, and 1 mM EDTA. Gels were dried and autoradiographed with intensifying screens at −70°C.

Antibodies

Monoclonal IgG antibody against α-spectrin (Affiniti, Exeter, UK) was used to assess calpain as well as caspase-3 activity. Bax and Bcl-2 monoclonal IgG antibodies (Santa Cruz Biotech, Santa Cruz, CA, USA) were used to assess apoptotic threshold by determining the Bax:Bcl-2 ratio. We have also used the IgG antibody against COX4, which is a membrane protein in the inner mitochondrial membrane and it remains in the mitochondria regardless of activation of apoptosis [23]. All other IgG antibodies were purchased either from Calbiochem (San Diego, CA, USA) or Santa Cruz Biotechnology (Santa Cruz, CA, USA). Monoclonal IgG antibody against β-actin (clone AC-15, Sigma Chemical Co., St. Louis, MO, USA) was used to standardize protein loading on the SDS-PAGE gels. We used horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG secondary antibody (ICN Biomedicals, Aurora, OH, USA) for detecting all primary antibodies against specific proteins, except calpain and α-spectrin where we used HRP-conjugated goat anti-rabbit IgG secondary antibody (ICN Biomedicals).

Measurement of Nitrite

Production of nitric oxide (NO) from cells was determined by measuring the amount of nitrite, a stable oxidation product of NO, as described previously [24]. An aliquot of the conditioned medium was mixed with an equal volume of 1% sulfanilamide in water and 0.1% N-1-naphthyle-thylenediamine dihydrochloride in 5% phosphoric acid. The absorbance was determined at 550 nm. Sodium nitrite, which was dissolved and diluted in culture medium for concentrations of 1–100 μM, was used to generate a standard curve.

Analysis of Levels of mRNA Expression by Reverse Transcription (RT)-PCR and Agarose Gel Electrophoresis

Extraction of total RNA and RT-PCR were performed according to standard procedure, as described previously [25]. All human primers (Table 1) for RT-PCR experiments were designed using Oligo software (National Biosciences, Plymouth, MN, USA) and custom synthesized (Operon Technologies, Alameda, CA, USA). The level of β-actin gene expression served as an internal control. The RT-PCR products were resolved on the agarose gels, stained with ethidium bromide (1 μg/ml), examined on a UV (303 nm) transilluminator, and protographed. We inverted gel pictures to present the RT-PCR bands in black. Quantity One software (Bio-Rad, Hercules, CA, USA) was used to measure optical density (OD) of each band. The mRNA expression of a target gene was normalized to that of β-actin.

Table 1.

Human primers used for determining levels of mRNA expression of specific genes

Gene	Primer sequences	Product size (bp)
β-actin	Sense: 5′-GTG GGG CGC CCC AGG CAC CA-3′	436
	Antisense: 5′-CTC CTT AAT GTC ACG CAC GAT TTC-3′
BIRC-2	Sense: 5′-CAG AAA GGA GTC TTG CTC GTG-3′	536
	Antisense: 5′-CCG GTG TTC TGA CAT AGC ATC-3′
BIRC-3	Sense: 5′-GGG AAC CGA AGG ATA ATG CT-3′	368
	Antisense: 5′-ACT GGC TTG AAC TTG ACG GAT-3′
BIRC-4	Sense: 5′-AAT GCT GCT TTG GAT GAC CTG-3′	470
	Antisense: 5′-ACC TGT ACT CAG CAG GTA CTG-3′
BIRC-5	Sense: 5′-GCC CCA CTG AGA ACG-3′	302
	Antisense: 5′-CCA GAG GCC TCA ATC C-3′
BIRC-6	Sense: 5′-AGC CGA AGG ATA GCG A-3′	385
	Antisense: 5′-GCC ATC CGC CTT AGA A-3′
BIRC-7	Sense: 5′-GCC TCC TTC TAT GAC T-3′	283
	Antisense: 5′-CGT CTT CCG GTT CT-3′
BIRC-8	Sense: 5′-GTG AGC GCT CAG AAA GAC ACT AC-3′	209
	Antisense: 5′-CAC ATG GGA CAT CTG TCA ACT G-3′
GFAP	Sense: 5′-CGC CTC GAT CAA CTC A-3′	210
	Antisense: 5′-CTC CTC CAG CGA CTC AAT-3′
hTERT	Sense: 5′-GTA CAT GCG ACA GTT C-3′	418
	Sense: 5′-TTC TAC AGG GAA GTT CAC-3′
hTER	Sense: 5′-TCT AAC CCT AAC TGA GAA GGG CGT AG-3′	126
	Sense: 5′-GTT TGC TCT AGA ATG AAC GGT GGA AG-3′

Western Blotting

Western blotting was performed according to standard procedure, as we described previously [12–14]. We used the enhanced chemiluminescence (ECL) reagents from Amersham Pharmacia (Buckinghamshire, UK) and X-OMAT AR films from Eastman Kodak (Rochester, NY, USA) for detection of protein bands on the Western blots. The films were scanned on a UMAX PowerLook Scanner (UMAX Technologies, Fremont, CA, USA) using Photoshop software (Adobe Systems, Seattle, WA, USA), and OD of each band was determined using Quantity One software (Bio-Rad).

Preparation of Nuclear, Mitochondrial, and Cytosolic Protein Fractions for Western Blotting

Preparations of nuclear, cytoplasmic and mitochondrial fractions were performed according to standard procedures [13]. We used these fractions for Western blotting for cytochrome c, CAD, and STAT-1.

Colorimetric Assay for the Measurement of Caspase-3, Caspase-8, and Caspase-9 Activities

Measurements of caspase activities in cells were performed with the commercially available caspase-3, caspase-8, and caspase-9 assay kits (Sigma). Proteolytic activities of caspase-3, caspase-8, and caspase-9 were assayed colorimetrically on the basis of release of p-nitroanilide (pNA) from DEVD-pNA, IETD-pNA and LEHD-pNA peptides. The solution containing p-NA appears yellowish and has a high absorbance at 405 nm (∈_mM = 10.5). The concentration of the pNA released from the substrate was calculated from the absorbance values at 405 nm. Experiments were performed in triplicates.

Use of Specific Protease Inhibitors for Prevention of Cell Death

Cells were pretreated (1 h) with 10 μM caspase-1 inhibitor, z-Leu-Nle-CHO (calpeptin), caspase-9 inhibitor, or caspase-3 inhibitor IV (Calbiochem) at 37°C. Control cultures were pretreated (1 h) with an equivalent amount (≤0.1%) of DMSO or were left untreated. Cells were then exposed to IFN-γ and washed with ice-cold PBS. Cell death based on loss of membrane integrity was determined by the inability of cells to exclude the vital dye trypan blue. At the appropriate time interval, cells were removed from each treatment, diluted (1:1) with trypan blue, and counted (at least 500 cells were counted from each sample). Light microscopy was used to examine the cells and calculate the percent viability (cells capable of excluding the dye).

Statistical Analysis

Results were analyzed using StatView software (Abacus Concepts, Berkeley, CA, USA) and compared using oneway analysis of variance (ANOVA) with Fisher’s post hoc test. Data were presented as mean ± standard error of mean (SEM) of separate experiments (n ≥ 3). Significant difference between control and treated cells was indicated by *P < 0.01 or **P < 0.001. Significant difference between treated cells without and with protease inhibitor pretreatment was indicated by #P < 0.05.

Results

Retinoids Induce Differentiation with Upregulation of GFAP and Down Regulation of Telomerase

Treatment with ATRA or 13-CRA induced astrocytic differentiation with upregulation of GFAP and caused a decline in telomerase expression and activity in both T98G and U87MG cells (Fig. 1). Morphological features of astrocytic differentiation appeared clearly following treatment of glioblastoma cells with a retinoid (Fig. 1, panel a). Expression of GFAP, an astrocyte-specific intermediate filament protein, is a biochemical marker of astrocytic differentiation in gliomblatomas [12, 26]. Our results showed that the expression of GFAP was remarkably increased (2-fold) in differentiated cells, compared with control T98G and U87MG cells (Fig. 1, panels b and c). However, the precise molecular mechanism responsible for the upregulation of GFAP during astrocytic differentiation is uncertain. Since GFAP is considered a determining factor for astrocytic cell shape, the morphological alterations observed might have been mediated through the upregulation of GFAP. Expression of hTERT, the human telomerase catalytic subunit, was examined by RT-PCR (Fig. 1, panel d) and also by Western blotting (Fig. 1, panel e). The mRNA expression of hTER, the human telomerase RNA component, apparently did not change due to treatments. Telomerase activity was evaluated by the TRAP assay (Fig. 1, panel f). Our results demonstrated that inhibition of telomerase expression and activity was associated with induction of differentiation in glioblastoma cells. These findings suggested that ATRA or 13-CRA induced astrocytic differentiation with overexpression of GFAP and down regulation of telomerase expression and activity in T98G and U87MG cells.

Fig. 1.

Induction of differentiation with upregulation of GFAP and down regulation of telomerase expression and activity in T98G and U87MG cells. (a) Methylene blue staining showed morphological features of astrocytic differentiation. Treatments for 7 days: control (CTL), 1 μM ATRA, and 1 μM 13-CRA. (b) Examination of GFAP expression at mRNA level by RT-PCR. (c) Examination of GFAP expression at protein level by Western blotting. (d) Examination of telomerase (hTERT) expression by RT-PCR. (e) Examination of telomerase (hTERT) expression by Western blotting. (f) TRAP assay to examine telomerase activity in cells. Treatments (panels b–f): CTL, 200 units/ml IFN-γ (48 h), 1 μM ATRA (7 days), 1 μM ATRA (7 days) + 200 units/ml IFN-γ (48 h), 1 μM 13-CRA (7 days), and 1 μM 13-CRA (7 days) + 200 units/ml IFN-γ (48 h). Marker lane (in panel f) contained a 20-bp DNA ladder (Gene Choice, Frederick, MD, USA)

Morphological and Biochemical Features of Apoptosis

The amount of apoptotic cell death was determined using Wright staining and ApopTag assay (Fig. 2). Treatment with ATRA or 13-CRA did not induce apoptosis whereas IFN-γ alone and especially combination of a retinoid and IFN-γ produced morphological features of apoptosis in both T98G and U87MG cells, as revealed by Wright staining (Fig. 2, panel a). These results from Wright staining were further confirmed by ApopTag assay that labeled DNA fragmentation as brown color in apoptotic cells (Fig. 2, panel b). Control cells and cells treated with ATRA or 13-CRA showed no brown color confirming almost absence of apoptosis. But cells treated with IFN-γ alone and combination of a retinoid and IFN-γ demonstrated prominent brown color apoptotic cells. Importantly, combination of a retinoid and IFN-γ produced the highest amount of apoptosis in both T98G and U87MG cells, as determined on the basis of ApopTag assay (Fig. 2, panel c). Thus, Wright staining and ApopTag assay clearly demonstrated apoptotic features morphologically and biochemically, respectively, in glioblastoma cells.

Fig. 2.

Morphological and biochemical features of apoptosis in T98G and U87MG cells. (a) Wright staining for morphological features of apoptosis. (b) ApopTag assay for labeling of DNA fragmentation in apoptotic cells. (c) Determination of amounts of apoptosis. Treatments (panels a–c): CTL, 200 units/ml IFN-γ (48 h), 1 μM ATRA (7 days),1 μM ATRA (7 days) + 200 units/ml IFN-γ (48 h), 1 μM 13-CRA (7 days), and 1 μM 13-CRA (7 days) + 200 units/ml IFN-γ (48 h)

Retinoids Reduced Inflammatory Response and Decreased Levels of Inducible Nitric Oxide Synthase (iNOS) and Production of Nitric Oxide (NO)

Although the anti-inflammatory effect of retinoid has been investigated for several decades, the underlying mechanisms responsible for this effect are largely unknown. In this study, we demonstrated that pretreatment with a retinoid inhibited IFN-γ mediated inflammatory responses in T98G and U87MG cells (Fig. 3). Both ATRA and 13-CRA decreased IFN-γ mediated expression of the pro-inflammatory molecule nuclear factor kappa B (NFκB) in glioblastoma cells (Fig. 3, panel a). These results showed that ATRA and 13-CRA induced an anti-inflammatory effect by suppressing the activation of the NFκB in IFN-γ treated gliomblastoma cells. In order to characterize the biochemical pathway responsible for pro-inflammatory effect of IFN-γ in glioblastoma cells, we investigated down stream signaling events. It has been shown in other cellular systems that IFN-γ can signal via STAT-1 [27]. Phosphorylation of STAT-1 (p-STAT-1) makes the molecule active and promotes its homodimerization for translocation to the nucleus and binding to the STAT1-specific DNA sequence. We detected acccumulation of p-STAT-1 in the nucleus of the cells treated with IFN-γ alone for 48 h (Fig. 3, panel a). In order to substantiate that nuclear accumulation of p-STAT-1 was due to signaling events such as phosphorylation and translocation, we used a monoclonal antibody that could recognize only the p-STAT-1 in nuclear fraction. Western blotting for nuclear p-STAT-1 levels showed that cells not treated with IFN-γ did not have nuclear accumulation of p-STAT-1 (Fig. 3, panel a). However, the highest induction and nuclear accumulation of p-STAT-1 occurred upon IFN-γ stimulation for 24 h (data not shown). Pretreatment of the cells with ATRA or 13-CRA substantially reduced the expression of NFκB (Fig. 3, panel a). Later we found that the DNA sequence with GAS element could bind specifically to p-STAT-1, which was induced by IFN-γ in T98G and U87MG cells (Fig. 3, panel b). Treatment of cells with IFN-γ induced p-STAT-1 for binding to GAS element for induction of the IFN-γ responsive genes. Thus, STAT-1 signaling pathway provides at least one mechanism that IFN-γ uses in T98G and U87MG cells for pro-inflammatory response. These data for the first time demonstrated a role for STAT-1 in human glioblastoma T98G and U87MG cells.

Fig. 3.

Retinoids reduced levels NFκB, iNOS, and NO in T98G and U87MG cells. (a) The representative Western blots showing levels of NFκB, STAT-1, p-STAT-1, and β-actin. (b) EMSA for examination of formation of p-STAT-1 and GAS complex. (c) The representative Western blots showing levels of iNOS and β-actin. (d) Determination of production of NO. Treatments (panels a, c, and d): CTL, 200 units/ml IFN-γ (48 h), 1 μM ATRA (7 days),1 μM ATRA (7 days) + 200 units/ml IFN-γ (48 h), 1 μM 13-CRA (7 days), and 1 μM 13-CRA (7 days) + 200 units/ml IFN-γ (48 h)

Production of NO through the iNOS pathway is increased in inflammatory diseases and cancers [24]. Our results showed that retinoids attenuated the levels of expression of iNOS (Fig. 3, panel c) and therefore reduced production of NO in glioblastoma cells (Fig. 3, panel d). Retinoids inhibited expression of iNOS by using the molecular mechanism that might involve inhibition of NFκB expression as well. Thus, retinoids attenuated inflammatory response to favor IFN-γ mediated apoptosis in glioblastoma cells.

IFN-γ Treatment Increased Caspase-1 Expression

Caspase-1 plays a key role in the processing of cytokines for apoptosis in neurons and macrophages [28]. It remains unclear whether caspase-1 also causes apoptosis in cancer cells. Our Western blotting showed an increase in caspase-1 expression in T98G and U87MG cells following treatment with IFN-γ alone and combination of retinoid and IFN-γ (Fig. 4, panel a). Our data also showed that pretreatment of cells with caspase-1 inhibitor I could significantly reduce IFN-γ induced cell death (Fig. 4, panel b). In general, caspase-1 has not been considered relevant to apoptosis in cancer. However, our data suggested that caspase-1 could play a role for induction of apoptosis in gliblastoma cells.

Fig. 4.

Activation of caspase-1 for induction cell death in T98G and U87MG cells. (a) The representative Western blots showing active caspase-1 and β-actin. (b) Determination of residual cell viability by trypan blue dye exclusion test and effect of inhibition of caspase-1 on residual cell viability. Treatments (panels a and b): CTL, 200 units/ml IFN-γ (48 h), 1 μM ATRA (7 days),1 μM ATRA (7 days) + 200 units/ml IFN-γ (48 h), 1 μM 13-CRA (7 days), and 1 μM 13-CRA (7 days) + 200 units/ml IFN-γ (48 h). In panel b, some treatments also included use of 10 μM caspase-1 inhibitor

Apoptosis with Activation of Caspase-8 and Proteolytic Cleavage of Bid to tBid

Caspase-8 is activated upstream of caspase-3 and recently it has been implicated to play a prominent role in induction of cell death [29]. Upon activation, caspase-8 uses Bid as a substrate to make truncated Bid (tBid) that may translocate to the mitochondria to induce cell death. We examined activation of caspase-8 for formation of tBid that could be translocated to mitochondia to trigger apoptosis (Fig. 5). Our results showed that treatment of cells with IFN-γ alone and combination of retinoid (ATRA or 13-CRA) and IFN-γ could generate caspase-8 active fragment for proteolytic cleavage of Bid to tBid, which was translocated to mitochondria (Fig. 5, panel a). After translocation from cytosol to mitochondria, tBid stimulates efficient oligomerization of Bax to activate the intrinsic pathway of apoptosis [30]. We measured the mitochondrial fraction for Wester blot analysis of tBid. β-Actin expression was used to ensure that equal amount of protein was loaded in each lane (Fig. 5, panel a). Further, we used a colorimetric assay to measure total caspase-8 activity in apoptosis of glioblastoma cells (Fig. 5, panel b). Treatment with the combination of a retinoid (ATRA or 13-CRA) and IFN-γ induced the highest caspase-8 activity for apoptosis in both T98G and U87MG cells.

Fig. 5.

Death receptor-dependent apoptosis with increases in caspase-8 activation and activity in T98G and U87MG cells. (a) The representative Western blots showing levels of active caspase-8 fragments in cytosol, β-actin in cytosol, formation and translocation of tBid to mitochondria, and COX4 in mitochondria. (b) Determination of caspase-8 activity using a colorimetric assay kit. Treatments (panels a and b): CTL, 200 units/ml IFN-γ (48 h), 1 μM ATRA (7 days), 1 μM ATRA (7 days) + 200 units/ml IFN-γ (48 h), 1 μM 13-CRA (7 days), and 1 μM 13-CRA (7 days) + 200 units/ml IFN-γ (48 h)

Apoptosis with Involvement of Mitochondria-Dependent Pathway

Members of the Bcl-2 family proteins play a major role in governing the mitochondria-dependent pathway of apoptosis, with proteins such as Bax functioning as inducers of apoptosis and proteins such as Bcl-2 as suppressors of apoptosis. We examined the involvement of mitochondria-dependent pathway of apoptosis in glioblastoma cells following treatments (Fig. 6). Our results showed an increase in the expression of Bax (pro-apoptotic protein) in cells treated with IFN-γ alone and combination of retinoid (ATRA or 13-CRA) and IFN-γ (Fig. 6, panel a). We performed densitometric analysis to determine the Bax:Bcl-2 ratio in all treatment groups and found that increases in Bax:Bcl-2 ratio correlated clearly with a commitment to apoptosis in cells treated with IFN-γ alone and combination of retinoid and IFN-γ (Fig. 6, panel b).

Fig. 6.

Mitochondria-dependent apoptosis with increases Bax:Bcl-2 ratio, mitochondrial release of cytochrome c, and caspase-9 activation and activity in T98G and U87MG cells. (a) The representative Western blots showing protein levels of Bax, Bcl-2, and β-actin. (b) Densitometric analysis showing the Bax:Bcl-2 ratio. (c) The representative Western blots showing protein levels of cytochrome c, COX4, caspase-9, and β-actin. (d) Determination of caspase-9 activity using colorimetric assay kit. Treatments (panels a to d): CTL, 200 units/ml IFN-γ (48 h), 1 μM ATRA (7 days), 1 μM ATRA (7 days) + 200 units/ml IFN-γ (48 h), 1 μM 13-CRA (7 days), and 1 μM 13-CRA (7 days) + 200 units/ml IFN-γ (48 h)

Any treatment of cells involving IFN-γ promoted disappearance 15 kD cytochrome c from the mitochondrial fraction (Fig. 6, panel c), indicating that IFN-γ induced mitochondrial release of cytochrome c. Because of release from mitochondria, 15 kD cytochrome c appeared in the cytosolic fractions (Fig. 6, panel c). We used COX4 as a loading control for mitochondrial fractions. Thus, apoptosis required the release of cytochrome c from mitochondria into the cytosol to cause activation of caspases. We found an increase in levels of 37 kD caspase-9 active fragment in T98G and U87MG cells undergoing apoptosis. Moreover, we found significant increases in total caspase-9 activity (colorimetric assay) in course of apoptosis (Fig. 6, panel d). These results suggested that caspase-9 activation might be a consequence of cytochrome c release from mitochondria. β-Actin was used to ensure that equal amount of protein was loaded in each lane. Taken together, the results showed an increase in Bax:Bcl-2 ratio, release of cytochrome c from mitochondria, and subsequent increases in caspase-9 activation and activity, indicating the involvement of mitochondria-dependent pathway of apoptosis in glioblastoma cells.

Apoptosis with Mitochondrial Release of Smac that Could Suppress BIRC Expression

The BIRCs and their counteraction by the mitochondrial protein Smac have also emerged as important regulators of mitochondria-dependent caspase activation and apoptosis in various cell systems. Like cytochrome c, Smac is located in mitochondria and released into the cytosol when cells undergo apoptosis. In response to apoptotic stimuli, Smac is released into the cytosol to promote caspase activation by binding to BIRCs for blocking their function [31]. These observations have suggested that Smac is an important regulator of apoptosis. To better understand the role of Smac in T98G and U87MG cells cells, we examined mitochondrial release of Smac into the cytosol and regulation of BIRCs in response to IFN-γ treatment (Fig. 7). Our results indicated that IFN-γ induced mitochondrial release of Smac for its appearance in the cytosolic fraction (Fig. 7, panel a). Treatment of the cells with the combination of a retinoid and IFN-γ caused the highest accumulation of Smac in the cytosol. Then, we studied whether expression of BIRCs was altered in T98G and U87MG cells following treatments (Fig. 7, panel b). The mRNA levels of BIRC-2 to BIRC-8 were assessed by RT-PCR experiments (Fig. 7, panel b) and the protein levels of BIRC-2 to BIRC-5 were examined by Western blotting (Fig. 7, panel c). The reduction of expression of BIRC-2 to BIRC-5 at both mRNA and protein levels correlated well with the Smac release from the mitochondria. The mRNA level of BIRC-6 was also decreased for favoring apoptotic death. However, levels of BIRC-7 and BIRC-8 remained relatively unchanged in the cells after any treatment involving IFN-γ, indicating that BIRC-7 and BIRC-8 did not play a major role in IFN-γ mediated apoptosis. These results indicated that IFN-γ induced apoptosis with mitochondrial release of Smac into the cytosol and decrease in levels of some BIRCs in T98G and U87MG cells (Fig. 7).

Fig. 7.

Mitochondrial release of Smac into the cytosol and suppression of BIRCs in T98G and U87MG cells. (a) The representative Western blots showing protein levels of Smac, COX4, and β -actin. (b) The representative RT-PCR gel pictures showing mRNA levels of BIRC-2 to BIRC-8 and β-actin. (c) The representative Western blots showing protein levels of BIRC-2 to BIRC-5 and β-actin. Treatments (panels a to c): CTL, 200 units/ml IFN-γ (48 h), 1 μM ATRA (7 days), 1 μM ATRA (7 days) + 200 units/ml IFN-γ (48 h), 1 μM 13-CRA (7 days), and 1 μM 13-CRA (7 days) + 200 units/ml IFN-γ (48 h)

Apoptosis Occurred with Increase in Intracellular Free [Ca²⁺], Activation of Calpain and Caspase-3, and Down Regulation of Calpastatin

We examined activation and activity of calpain (a Ca²⁺-dependent cysteine protease) and caspase-3 in course of apoptosis in glioblastoma cells after the treatments (Fig. 8). We used fura-2 assay to determine the changes in intracellular free [Ca²⁺] in T98G and U87MG cells (Fig. 8, panel a). Treatment of cells involving IFN-γ caused a significant increase in intracellular free [Ca²⁺] in cells, suggesting a role for Ca²⁺ influx for apoptosis in T98G and U87MG cells. Western blotting showed an increase in 76 kD calpain active fragment and 20 kD caspase-3 active fragment in apoptotic cells (Fig. 8, panel b). Calpain plays a dual role, mediation of Ca²⁺ influx and proteolysis subsequent to Ca²⁺ influx, during cell death. The degradation of 270 kD α-spectrin to 145 kD spectrin breakdown product (SBDP) and 120 kD SBDP has been attributed to activities of calpain and caspase-3, respectively [12–14]. So, we examined the activities calpain and caspase-3 in the formation of calpain-specific 145 kD SBDP and caspase-3-specific 120 kD SBDP, respectively, on the Western blots. Cells treated with IFN-γ showed increases in 145 kD SBDP and 120 kD SBDP. Levels of 100 kD calpastatin, an endogenous calpain inhibitor and also a calpain substrate, were decreased during apoptosis in T98G and U87MG cells (Fig. 8, panel b). It has previously been reported that this high-molecular-weight form of calpastatin can be extensively degraded in course of apoptosis [32].

Fig. 8.

Increases in intracellular free [Ca²⁺] and activation of calpain and caspase-3 in T98G and U87MG cells. (a) Determination of percent increase in intracellular free [Ca²⁺] using fura-2 assay. (b) The representative Western blots showing levels of calpain, calpastain, caspase-3, SBDP, ICAD, CAD, and β-actin. (c) Determination of caspase-3 activity using a colorimetric assay kit. Treatments (panels a–c): CTL, 200 units/ml IFN-γ (48 h), 1 μM ATRA (7 days), 1 μM ATRA (7 days) + 200 units/ml IFN-γ (48 h), 1 μM 13-CRA (7 days), and 1 μM 13-CRA (7 days) + 200 units/ml IFN-γ (48 h)

Increase in activation and activity of caspase-3 can release caspase-3-activated DNase (CAD) due to cleavage of inhibitor of CAD (ICAD) in the cytosol [33]. Then, CAD could enter the nucleus to degrade the chromosomal DNA. Our results showed decrease in levels of 45 kD ICAD in the cytosolic fractions and increase in levels of 40 kD CAD in the nuclear fractions in both T98G and U87MG cells after treatments involving IFN-γ (Fig. 8, panel b). For analysis of levels of nuclear CAD, we ran two sets of SDS-PAGE gels at the same time. One set of gels with resolved proteins was used for analyzing levels of nuclear CAD by Western blotting; and other set of SDS-PAGE gels was stained with Coomassie Blue to confirm equal amounts of nuclear protein loading in all lanes (Fig. 8, panel b). We also used colorimetric assay to determine total caspase-3 activity, which was significantly increased in cells treated with IFN-γ alone or combination of retinoid and IFN-γ (Fig. 8, panel c).

Use of Inhibitor Showed the Involvement of Proteases in Apoptosis

We examined the extent of contribution of specific proteases in the process of cell death (Fig. 9). Pretreatment of cells for 1 h with 10 μM caspase-8 inhibitor II, calpeptin (calpain inhibitor), caspase-9 inhibitor I, or caspase-3 inhibitor IV significantly inhibited cell death by IFN-γ, indicating major contribution of these proteases in cell death (Fig. 9).

Fig. 9.

Inhibition of specific proteases indicated their roles in cell death in T98G and U87MG cells. Trypan blue dye exclusion test for determination of residual cell viability after inhibition of (a) caspase-8, (b) calpain, (c) caspase-9, and (d) caspase-3 activities by their specific inhibitors. Treatments (panels a to d): CTL, 200 units/ml IFN-γ (48 h), 1 μM ATRA (7 days), 1 μM ATRA (7 days) + 200 units/ml IFN-γ (48 h), 1 μM 13-CRA (7 days), and 1 μM 13-CRA (7 days) + 200 units/ml IFN-γ (48 h). As shown, cells were also pretreated (1 h) with 10 μM of (a) caspase-8 inhibitor II, (b) calpeptin (calpain inhibitor), (c) caspase-9 inhibitor I, and (d) caspase-3 inhibitor IV

Based on the experimental results, we suggested that IFN-γ induced apoptosis in glioblastoma cells via both death receptor and mitochondria mediated pathways (Fig. 10).

Fig. 10.

A schematic presentation of molecular mechanisms leading to induction of differentiation and enhancement of apoptosis with activation of calpain and caspase cascades in human glioblastoma T98G and U87MG cells following treatment with combination of a retinoid (ATRA or 13-CRA) and IFN-γ. An arrow indicates pathway moving forward or upregulation of a component while a bar indicates pathway blocked

Discussion

Chemotherapy using a single agent may be limited because of toxicity and lack of potency. Therefore, combination chemotherapeutic strategies are currently being extensively explored to reduce toxicity and enhance chemotherapeutic actions. The combination of a differentiating agent (e.g., ATRA or 13-CRA) with a cytokine (e.g. IFN-γ) may be beneficial and relevant for controlling the growth of glioblastoma. Induction of differentiation followed by activation of apoptosis in malignant cells has been considered a promising strategy in cancer therapy. Our results imply that retinoid induces differentiation as well as reduces inflammatory process to increase sensitivity of human glioblastoma cells to IFN-γ for apoptosis. In this investigation, we studied the molecular mechanisms that relate to induction of differentiation and reduction of inflammation for enhancement of apoptosis in T98G and U87MG cells following treatment with the combination of a retinoid (ATRA or 13-CRA) and IFN-γ (Figs. 1–10).

Our results provided direct evidence for overexpression of GFAP and down regulation of telomerase in differentiated T98G and U87MG cells (Fig. 1). GFAP is an astrocyte-specific intermediate filament that provides structural support to normal astrocytes. The induction of overexpression of GFAP in astrocytomas is indicative of a more differentiated astrocytic phenotype [34].

Our data also showed an increase in apoptotic cell death in T98G and U87MG cells treated with combination of a retinoid and IFN-γ (Fig. 2). Inhibition of iNOS could be an effective anti-inflammatory strategy. The present study explored whether retinoids, known to have anti-inflammatory actions, could attenuate cytokine-stimulated iNOS expression in glioblastoma cells. Indeed, treatment with a retinoid reduced production of inflammatory factor (NFκB), iNOS, and NO to enhance sensitivity of glioblastoma cells to IFN-γ for apoptosis (Fig. 3). These results indicated that differentiation attenuated inflammation and enhanced sensitivity to IFN-γ. We also demonstrated that putative GAS element could bind specifically to a cellular factor (p-STAT-1), which was induced by IFN-γ in T98G and U87MG cells. Activation of STAT-1 signaling pathway by IFN-γ provides at least one mechanism that mediates induction of apoptosis in T98G and U87MG cells.

Cells treated with IFN-γ showed an increase in caspase-1 expression (Fig. 4). Caspase-1 is categorized as a cytokine-processing caspase. Our results suggest that caspase-1 can contribute to induction of apoptosis in glioblatoma cells and pretreatment of cells caspase-1 inhibitor inhibits cell death (Fig. 4). These results indicate that caspase-1 is a pro-apoptotic protease in glioblastoma cells. During apoptosis in glioblastoma cells, caspase-8 activity was increased followed by cleavage of Bid to tBid (Fig. 5). Caspase-8 mediated cleavage of the Bid provides a linkage between the death receptor and mitochondrial pathways of apoptosis. Our data also showed an increase in Bax:Bcl-2 ratio, release of cytochrome c from mitochondria, and increase in activity of caspase-9 for mediation of apoptosis (Fig. 6). The release of Smac from mitochondria also correlated clearly with the down regulation of BIRC-2 to BIRC-5 at mRNA and protein levels (Fig. 7). An increase in intracellular free [Ca²⁺] triggered activation of calpain and caspase-3 for degradation of calpastatin to promote apoptosis (Fig. 8). Our group previously reported that IFN-γ induced expression of calpain at the mRNA and protein levels in U937 and THP-1 cells [35]. Involvement of different proteolytic mechanisms in cell death were confirmed by pretreating the cells with caspase-1 inhibitor (Fig. 4), caspase-8 inhibitor II, calpeptin (calpain inhibitor), caspase-9 inhibitor I, and caspase-3 inhibitor IV (Fig. 9). Taken together, these data suggested that IFN-γ induced apoptosis by both death receptor and mitochondrial pathways.

A number of pro-apoptotic and anti-apoptotic members of the Bcl-2 protein family regulate the release of cytochrome c and Smac from the mitochondrial intermembrane space into the cytosol [36]. Cytochrome c then interacts with pro-caspase-9 and Apaf-1 to activate caspase-9 and thus switch on caspase-3 leading to apoptosis [36]. We confirmed the involvement of mitochondria in apoptosis in glioblastoma cells (Figs. 5–8). Our results showed that appearance of tBid preceded or coincided with the release of cytochrome c and the activation of caspases-9 and casapase-3 in T98G and U87MG cells.

Our observation that treatment with IFN-γ leads to an increase in calpain and caspase-3 activities confirms occurrence of apoptotic death in T98G and U87MG cells. Pro-apoptotic Bax is thought to be upstream of the cysteine proteases in the mitochondria-dependent pathway of apoptosis [37]. Our results confirmed this notion and showed that increase in caspase-3 activity could cleave ICAD to release and translocate CAD to the nucleus (Fig. 8) to cause chromosomal DNA fragmentation. Using different protease inhibitors, we confirmed that treatment of T98G and U87MG cells with IFN-γ induced apoptosis by activation of multiple proteolytic pathways (Fig. 9).

In conclusion, our results demonstrated that ATRA or 13-CRA down regulated telomerase activity and inflammation to enhance the efficacy of IFN-γ for increasing apoptosis in human glioblastoma T98G and U87MG cells (Fig. 10). The combination of a retinoid and IFN-γ caused more apoptosis than IFN-γ alone. Thus, our results provided the proof-of-principle for this novel combination approach that could further be tested in animal models of glioblastoma.